Michał Witkowski — Head of AI and Data at Fun Crafters
North Star metric, or how analytics can get your team to align and cooperate
September 20, 2024
9:00
Evidence-Based Management
🇬🇧 English
Evidence-Based Management
Metodyka projektu discovery
About
Dobre praktyki prowadzenia projektów odkrywczych. Przywództwo, cykl życia, zespół, innowator.
Watch the full talk
Watch this WaysConf session, then continue with related talks or explore the current programme.
From the recording
Talk in brief
In this WaysConf 2024 presentation, Marcin Żmigrodzki outlines a systematic methodology for managing discovery and exploration projects, drawing from his extensive research and experience in startup incubation and corporate innovation. He argues that discovery projects—which aim to acquire knowledge and validate assumptions as rapidly as possible—cannot be managed with traditional Waterfall or standard Agile frameworks due to their inherent chaos and non-linear timelines. Instead, successful discovery requires iterative governance loops, clear maturity models (such as NASA's Technology Readiness Levels or hypothesis-based business stages), and a distinct leadership style that balances extreme determination with team autonomy. Żmigrodzki highlights the fundamental psychological conflict between the researcher's mindset, which embraces uncertainty and delayed gratification, and the producer's mindset, which thrives on predictability and completed deliverables. Consequently, he expresses skepticism toward popular 'dual-track' approaches, recommending instead that organizations separate discovery and delivery into distinct phases or specialized teams to maintain focus and psychological safety.
Key takeaways
- 01
The Core Objective of Discovery
The primary goal of any discovery or exploration project is to acquire critical knowledge and validate assumptions as quickly as possible. This process helps organizations rapidly eliminate unviable paths and identify promising directions.
Watch from 1:22 - 02
Iterative Governance and Funding Decisions
Discovery projects should be managed in loops where sponsors provide a temporary credit of trust to test specific hypotheses. At the end of each cycle, sponsors must decide whether to fund the next stage, cancel the project, or grant a short extension to resolve remaining ambiguities.
Watch from 5:31 - 03
The Limitations of Standard Frameworks
Traditional frameworks like Waterfall or Scrum are ill-suited for the chaotic, highly specialized nature of discovery work. Instead of synchronized sprints, discovery tasks must progress at their own natural rhythms based on the specific competencies required.
Watch from 25:27 - 04
The Paradox of Discovery Leadership
Successful innovation leaders are rarely just supportive 'servant leaders'; they are often highly demanding, technically deep, and fiercely determined. They grant their teams high autonomy during execution but act autocratically to force rapid pivots when a path proves unviable.
Watch from 19:20 - 05
Mindset Incompatibility in Dual-Track Systems
Combining discovery and delivery within a single team is highly challenging due to conflicting psychological motivators. Researchers are driven by uncertainty and delayed gratification, whereas producers require predictability and the satisfaction of a completed product.
Watch from 37:58
Read edited transcript highlights
These concise notes were edited from automatic captions and checked against the talk structure. They are not a verbatim transcript.
The True Nature of Corporate Innovation Bubbles
In large corporations, innovation is often stifled by a cultural preference for safety, predictability, and rigid standards like Six Sigma, which are designed to minimize behavioral variance. To bypass this, successful organizations establish small, isolated 'innovation bubbles' or collaborate with external accelerators and hackathons. These small teams are tasked with rapidly testing high-risk, high-reward ideas. Because discovery projects rely heavily on specialized, narrow expertise, they cannot be scaled to hundreds of people like traditional Waterfall projects. Instead, organizations must run multiple small, independent teams in parallel to explore different directions without creating organizational gridlock.
Watch from 1:46Navigating the Three Outcomes of a Discovery Loop
The governance of discovery projects operates on a trust-based loop rather than rigid milestones. Sponsors—whether they are corporate committees, venture capitalists, or government agencies—provide a temporary credit of trust and resources for a set period, such as a quarter or a year, to validate a list of core assumptions. At the end of this loop, three decisions are possible: if the path shows promise, the team receives funding for the next stage; if the hypotheses are disproven and the path is a dead end, the project is canceled and a new idea is introduced; or, if the results are inconclusive but trust remains, the team is granted a shorter extension to gather more data.
Watch from 7:52The Demanding Traits of Successful Tech Founders
Modern management literature heavily promotes the concept of supportive servant leadership, but historical evidence from successful tech giants paints a different picture. Many founders who successfully scaled highly innovative companies were demanding, manipulative, and sometimes feared by their employees. However, they possessed two critical traits: absolute determination to succeed and deep technical expertise. These leaders did not merely delegate tasks from a corporate distance; they actively participated in the technical work. Their teams remained loyal not because the leader was easy to work with, but because they believed in the leader's vision and wanted to be part of something groundbreaking.
Watch from 12:20Why Discovery Projects Resist Standard Sprints
Traditional project management relies on predictable schedules, while Agile frameworks like Scrum enforce synchronized, equal-length sprints. Discovery projects reject both approaches because different tasks naturally progress at completely different speeds. For example, in an AI project, training a model might take only an hour if computing power is purchased, allowing for rapid iterations. Conversely, in medical or hardware projects, gathering clinical data or waiting for specialized components from overseas can take months. Forcing highly specialized experts into a uniform sprint cycle is counterproductive; instead, they must be allowed to work in their own chaotic rhythms.
Watch from 24:54The Skepticism Surrounding Dual-Track Agile
While the 'Dual-Track Agile' model is highly popular in product management, there is a strong argument against attempting to combine discovery and delivery within the same team simultaneously. It is incredibly difficult for individuals to switch between the mindset of a researcher and that of a producer. A researcher must embrace extreme uncertainty and accept delayed gratification, sometimes working for years without knowing if a solution exists. A producer, however, is motivated by the immediate satisfaction of completing a high-certainty task, like a programmer writing clean code or a mechanic fixing a car. Separating these roles into distinct phases or teams is often far more effective.
Watch from 34:58Building Psychological Safety to Unlock Creativity
True creativity and innovation cannot be forced through corporate posters or superficial training sessions. Psychological research indicates that creativity is deeply tied to a sense of security and risk tolerance. In an organizational context, this translates to psychological safety—the absolute certainty that an employee will not be ridiculed or penalized for proposing a wild idea or failing during an experiment. If team members fear negative social or professional consequences, they will stop sharing unconventional ideas, effectively killing the organization's innovative potential. Leaders must actively build this safety by encouraging experimentation and personally supporting those who take calculated risks.
Watch from 45:31Explore WaysConf 2026
